Online optimization-based predictive flight control using first-order methods

In fields such as aerospace or automotive, the use of classical control methods such as PID is still significant. The presence of constraints, however, impacts on the performance of these controllers that are usually designed to avoid constraint saturation. MPC techniques are the obvious alternative to handle constraint saturation and fully exploit the operative range of the system. Furthermore, MPC can be used as a fault-tolerant controller to handle actuator faults and control reallocation in a modular and systematic way.

In this dissertation, we rely on MPC techniques to handle constraints and actuator faults, motivated by an aerospace application (i.e., the longitudinal control of an Airbus passenger aircraft). The proposed fault-tolerant strategy strongly relies on the MPC capability of handling constraints and directly controlling each actuator independently. The advantages of MPC in terms of performance and fault tolerance, however, are shadowed by the computational requirements of this technique. The presence of an optimizer compromises the performance of the controller in terms of computation time and increases its requirements in terms of hardware and software. In this dissertation, we design MPC-tailored optimizers suitable for online optimization. The proposed contributions aim to bring MPC closer to actual implementation on the next generation of aircraft.